Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Machine Learning for Low Resource Settings

Sponsored · SiteGround - Reliable hosting with speed, security, and support you can count on. →

Machine Learning for Low Resource Settings

We discuss tools, techniques, and experiences that helps us discover how we can turn raw data into measurable improvements in outcomes.

Peter Lubell-Doughtie

October 31, 2018

Transcript

7 slides
  1. Peter Lubell-Doughtie CTO & Co-Founder Machine Learning for Low Resource

    Settings How can we turn raw data into measurable improvements in outcomes?
  2. • Improved outcomes ◦ Reduce preventable deaths ◦ Manage workforce

    • Have impact at scale ◦ Automation ◦ Standards for best practices ◦ Reduce time to implementation ◦ Reduce time to correction • Applies to problems we have ◦ Find fake and exceptional data ◦ Estimate missing data ◦ Triage data by risk ◦ Predict demand and trends Why Use Machine Learning?
  3. Opportunities • Significant much room for improvement • Willingness to

    experiment Challenges • Current digital data not collected with ML in mind • Data sovereignty often requires on-premises deployments Working with governments and NGOs
  4. Opportunities • Flexibility, sometimes due to lack of pre-existing solutions

    • State-of-the-art open source solutions Challenges • Use ML as a means not an end • Data consistency across different implementations Scaling Data Management Platforms

More Decks by Peter Lubell-Doughtie

Other Decks in Technology